|
(11) | EP 0 443 676 A1 |
| (12) | EUROPEAN PATENT APPLICATION |
|
|
|
|
|||||||||||||||||||||||||||
| (54) | Transmission system for digitised television images |
| (57) A series of images of a full motion video scene is divided into a number of sub-series
which, when interleaved, represent the original series. A ranking order increasing
magnitude is allocated to consecutive sub-series. Each image of a sub-series is converted
into an image data block, while images from one or more sub-series having a lower
order of precedence are considered for this conversion. A packet header indicating
the ranking order of the sub-series with which the corresponding image is associated
is added to each image data block. The display apparatus is adapted to select those image data blocks which have predetermined packet headers and to subject the image data blocks thus selected to a decoding process. The number of different packet headers which is selected determines the temporal resolution of the scene to be displayed. |
A. Background of the invention.
A(1) Field of the invention.
[0001] The invention generally relates to a method of transmitting a series of images of a full motion video scene in a digital format via some transmission medium. More particularly, said transmission medium is constituted by a compact disc-like record carrier.
[0002] The invention also relates to a display apparatus in which the transmitted images are processed and made suitable for display on a display screen; and to an optically readable record carrier on which said images are stored.
A(2) Description of the prior art.
[0003] More than fifteen years ago the firm of Philips marketed an optically readable record carrier on which audio signals as well as analog video signals were recorded. This record carrier was referred to as video long play (VLP) and supplemented the well-known audio long play (ALP). As compared with videotapes, such optically readable record carriers have the advantage that their quality does not deteriorate due to repeated use. However, as compared with video tapes they have the drawback that they cannot be rerecorded.
[0004] In the last ten years a completely new trend has developed, namely that of the optically readable audio record carriers generally known by the name of CD audio (Compact Disc audio). Due to its general acceptance and the ever increasing demand for integration of audio and video apparatus, a compact disc video has been created on which digitised audio signals as well as an analog video signal are present, which video signal corresponds to a full motion video scene having a duration of several minutes.
[0005] To increase this duration, the original analog video signal has been digitised. A full motion video scene is then considered as a finite series of images, for example, fifty or sixty occurring each second. Such an image comprises, for example 288 image lines with 352 pixels per line. By means of some sensibly chosen encoding algorithm each image is converted into an image data block comprising so much digital information that each pixel of the image can be reconstructed, with the possible inclusion of the information from other image data blocks. The encoding algorithm is chosen to be such that consecutive image data blocks comprise a minimum amount of redundant information. Since the length of each image data block (number of bits in this image data block) is thus very limited, a very large number of such image data blocks can be recorded on such a record carrier.
B. Object and summary of the invention.
[0006] The invention has for its object to contribute to the above-mentioned novel development in order to render said display apparatus financially accessible to a very wide public on the consumer market.
[0007] According to the invention the images of the series are subjected to a hierarchic encoding process in which the original series of images is considered as a number of interleaved sub-series having an increasing ranking order and in which images from sub-series having a lower ranking order are considered for encoding an image of a sub-series. In this way each image is converted into an image data block and a packet header indicating the ranking order of the sub-series with which the corresponding image is associated is added to each image data block.
[0008] The display apparatus is now adapted to receive all these image data blocks but to select only those blocks which have predetermined packet headers. Only image data blocks which are thus selected are subjected to a hierarchic decoding process in a video processing circuit so as to generate signals which are suitable for displaying the image on a display screen (for example, a display tube).
[0009] The invention will certainly be appreciated if the following aspect is considered. The costprice of a video processing circuit increases exponentially with the number of operations (additions, subtractions, etc.) which it can perform each second. If the rate of the images in the original series is equal to 50 Hz, this means that the video processing circuit must be capable of determining each second the three chrominance signals R, G and B from the transmitted information for approximately 5.10⁶ pixels. The number of operations which must thus be performed is so high that this can only be realised by means of a very "powerful" video processing circuit which is, however, so costly that display apparatus is financially accessible to a select group of consumers only.
[0010] According to the invention the display apparatus can make a selection from the presented image data packets so that only image data packets having predetermined packet headers are applied to the video processing circuit for further processing. This means that the video processing circuit only has to process a part of all available image data packets, for example, no more than half of them. It is true that this is at the expense of the image quality, but practice has proved that this quality is maintained at a sufficiently high level. It also means that the video processing circuit may be considerably less powerful, which renders its costprice and hence that of the display apparatus very favorable.
C. Brief description of the Figures.
[0011]
Fig. 1 shows diagrammatically a compact disc-like record carrier having a track and its division into packets;
Figs. 2 to 7 show some diagrams to explain the hierarchic encoding process;
Fig. 8 shows a sequence in which the image data blocks with different packet headers can be transmitted,
Fig. 9 shows diagrammatically the structure of a display apparatus according to the invention;
Fig. 10 shows diagrammatically another implementation of the hierarchic encoding process.
D. Explanation of the invention.
[0012] In Fig. 1 a part of the track on a compact disc-like record carrier is shown diagrammatically at A. A packet is present each time between two consecutive points a, b, c, d, e, etc. The structure of such a packet is shown diagrammatically at B in Fig. 1. It comprises, for example 2352 bytes and is divided into a packet header H comprising 24 bytes and a data field D comprising 2328 bytes.
[0013] The packet header H is further divided into a synchronisation field SNC of 12 bytes, an ordinal number field RF of four bytes and a service field SF of eight bytes. The synchronisation field SNC marks the start of a packet. It comprises one byte consisting exclusively of "0" bits, followed by 10 bytes consisting exclusively of "1" bits and finally again one byte consisting exclusively of "0" bits. The bytes in the ordinal number field RF indicate the ordinal number of the packet in the track. The service field SF indicates whether the packet is a video packet, an audio packet or a computer data packet.
[0014] The data field D is divided into data slots DS. These data slots of an audio packet are chosen to be such that a 16-bit audio word of a digital audio signal can be transmitted in each slot. The data slots of a video packet are chosen to be such that an 8-bit video word of a digitised video signal can be incorporated in each slot. These data slots also have a length of one byte for computer data packets.
[0015] As already stated in the foregoing, each image is considered as a matrix of 288*352 pixels P(i,k). In this case i(= 1, 2, 3, ... 288) is the ordinal number of the row and k (=1, 2, ... 352) is the ordinal number of the pixel on this row (column). The color of such a pixel is completely determined by an associated luminance value Y(i,k) and two color difference values U(i,k) and V(i,k). If these three values of each pixel were encoded with an eight-bit accuracy, approximately 130 video packets would be required for one image. However, this number can be reduced to 54 video packets without any deterioration of the image quality, namely by transmitting only one out of four color difference signals in one out of two image lines. In this case an image is thus completely defined by a 288*352 luminance matrix Y(i,k), a 144*88 color difference matrix U(r,s) and a 144*88 color difference matrix V(r,s), r = 1, 2, ... 144 and s = 1, 2, ... 88.
[0016] There are many encoding methods of further reducing the number of bits required to represent an image and hence the number of video packets required for each image. By way of example one such method will now be described in greater detail with reference to Fig. 2. In this Figure 2 the reference S₀ denotes a series of consecutive images B₁, B₂, ... B₁₂ of a full motion scene. The luminance matrix associated with the image Bn (n = 1, 2, ...) will be denoted by Yn(i,k) and the color difference matrices will be denoted by Un(r,s) and Vn(r,s), respectively. For each image Bn a prediction image Bn' is determined, comprising the prediction matrices Yn'(i,k), Un'(r,s) and Vn'(r,s) and, starting from these matrices a difference image DBn comprising the difference matrices DYn(i,k), DUn(r,s) and DVn(r,s) by difference formation of the image Bn and the prediction image Bn', or expressed mathematically:
i.e.:
The prediction image Bn' is obtained by determining a system of motion vectors Qn-1,n for the previous image Bn-1 and by shifting the individual pixels of this image Bn-1 in accordance with the associated motion vectors.
[0017] Since the dynamic range of the luminance and color difference values of the difference matrices is considerably smaller than that of the original matrices, these values can be represented with considerably fewer bits, for example with only four bits instead of the original eight bits. Although the calculated systems of motion vectors Qn must be transmitted in addition to the difference images DBn for an accurate reconstruction of the original images in the display apparatus, this method results in a considerable saving of bits. On the one hand a larger number of images can thus be recorded on the record carrier and on the other hand the time required to read all information for an image from the record carrier is considerably shorter.
[0018] In this known encoding method each difference image is dependent on the previous image. In the display apparatus each image of the series will therefore have to be reconstructed. This means that the temporal resolution of the scenes to be displayed by the display apparatus is equal to the temporal resolution of the scenes which have been picked up. As already noted, this means that the display apparatus should comprise a very powerful video processing circuit.
[0019] The temporal resolution can be influenced and hence the requirements which must be imposed on the video processing circuit can be influenced by subjecting the images of the series to a hierarchic encoding process as extensively described, for example in European Patent Application no. 0,340,843. For the sake of completeness this method will be described in greater detail by way of example with reference to Fig. 3. In this Figure 3 the series of consecutive images B₁, B₂, ... B₁₂ of a full motion scene is again shown at S₀. This series is divided into a number of sub-series, four in this case, denoted by S₁, S₂, S₃ and S₄, respectively.
Sub-series S₁ comprises the images B₁, B₅, B₉,...,
sub-series S₂ comprises the images B₃, B₇, B11, ...,
sub-series S₃ comprises the images B₂, B₆, B₁₀, ..., and
sub-series S₄ comprises the images B₄, B₈, B₁₂, ....
The images of sub-series S₁ are converted into difference images DB₁, DB₅, DB₉, ... in the way as described above with reference to Fig. 2. As is shown in Fig. 4 for the sake of completeness, a system of motion vectors is more particularly determined for each image of this sub-series S₁. The system Q1,5 for the image B₁, the system Q5,9 for the image B₅, the system Q9,13 for the image B₉ and so forth. With the aid of these vectors prediction images B₁', B₅ ', B₉', ... are calculated and the difference image DBm is obtained from a series DS₁ by difference formation of the original image Bm (m = 1, 5, 9, 13, ...) and the associated prediction image Bm'.
[0020] As already noted, a vector of, for example, the system Q1,5 denotes the direction and the distance over which a pixel or a group of pixels of the image B₁ must be displaced so as to reach the position of this pixel or group of pixels in the image B₅. For encoding the images in the sub-series S₂, S₃ and S₄ it is assumed for the sake of simplicity that such a displacement is linear. This means that said pixel of B₁ has undergone a quarter of the total displacement for the image B₂, half the total displacement for the image B₃ and three quarters of the total displacement for the image B₄. For encoding the images of the sub-series S₂ one proceeds in the manner as shown in Fig. 5. Starting from the image B₁ and a system of motion vectors ½ Q1,5 each having the same direction as the motion vectors in the system Q1,5, but being only half as long, a prediction image B1,3 is determined. Starting from the image B₅ and a system of motion vectors-½ Q1,5 each having a direction which is opposite to the direction of the motion vectors in the system Q1,5 and being only half as long, a prediction image B5,3 is determined. The average value of the two prediction images B1,3 and B5,3 is taken by adding the two prediction images together and dividing them by two. The result is the desired prediction image B₃'. By difference formation with the original image B₃, the difference image DB₃ of a series DS₂ is obtained.
[0021] As is shown in Fig. 6, a prediction image B₂' is determined in a corresponding manner, starting from the images B₁ and B₃, which prediction image leads to a difference image DB₂ of a series DS₃ by difference formation with B₂. Finally Fig. 7 shows how a difference image DB₄ of a series DS₄ is obtained by starting from the images B₃, B₄ and B₅.
[0022] For transmitting the series of sub-images thus obtained, the information for each sub-image is serialised so that an image data block for each sub-image is obtained. The image data block associated with the difference image DBn will be denoted by DB
. The image data blocks thus obtained are subsequently transmitted (i.e. recorded on the disc) in the sequence as shown, for example in Fig.8. More particularly, an image data block (for example, DB
) associated with a difference image from series DS₁ is transmitted first, then the image data block (DB
) of the immediately preceding difference image associated with series DS₂, subsequently the image data block (DB
) of the immediately preceding difference image associated with series DS₃ and finally the image data block (DS
) of the immediately preceding difference image associated with series DS₄. It is to be noted that B₁ in Fig. 8 is assumed to be the first image of the scene.
[0023] To be able to distinguish the image data blocks of the difference images of series DSi (i = 1, 2, 3, 4) from those of the difference images of series DSj (j = 1, 2, 3, 4,) and j ≠ i, a packet header indicating the series with which a corresponding difference image is associated is added to each image data block. In Fig. 8 these packet headers are denoted by DS₁, DS₂, DS₃ and DS₄.
[0024] Fig. 9 diagrammatically shows an embodiment of a display apparatus adapted to receive digitised images which are transmitted by means of a compact disc-like transmission medium in the format shown by way of example in Fig. 8. This display apparatus is provided with a read device 1 by means of which information recorded on a compact disc-like record carrier 2 can be read and converted into an electric signal which is applied to a demultiplexer 3. Starting from the information in the service field SF of a packet on the disc, this demultiplexer supplies the computer data packets at its output 3(1), the audio packets at its output 3(2) and the video packets at its output 3(3).
[0025] Since only the processing of the video packets plays a role within the scope of the present invention, the processing of the audio and computer data packets will not be further dealt with. The video packets are applied to a selection circuit 4 removing the packet headers from the video packets and selecting those blocks from the remaining image data blocks which are provided with predetermined packet headers, for example, only those image data blocks which are provided with the packet header DS₁, or both those image data blocks which are provided with the packet header DS₁ and those image data blocks which are provided with the packet header DS₂, etc. The image data blocks thus selected are applied to the video processing circuit 5 which supplies a luminance matrix Y(i,k) and the associated color difference matrices U(r,s) and V(r,s) for each image to be displayed. In the embodiment shown the luminance matrix Y(i,k) is stored in a luminance memory 6(1), the color difference matrix U(r,s) is stored in a U memory 6(2) and the color difference matrix V(r,s) is stored in a V memory 6(3). These memories 6 (.) are addressed in the conventional manner by addresses ADD of an address generator 7 and by a read-write enable signal R/W(.). As soon as this signal has the logic value "1", information can be written in the relevant memory. If it has the logic value "0", the contents of the memory can be read. The information read from a memory 6(.) is converted in a D/A converter 8(.) into an analog signal. The analog luminance signal Y(t) thus obtained, as well as the two analog color difference signals U(t) and V(t) are converted into the elementary chrominance signals R, G and B in a dematrixing circuit 9 and applied to a display tube 10.
[0026] It will be evident that the more powerful the video processing circuit 5 is (and consequently the more costly), the more series of difference images can be selected by the selection circuit 4 (number of different packet headers) and thus the higher the temporal resolution will be.
[0027] It has been tacitly assumed in Fig. 3 that the rate at which the images occur in the original series is equal to 50 Hz. However, the present invention obviates the ever recurrent problem related to the difference between the so-called 50 and 60 Hz field frequency countries. Let it be assumed that the images shown in Fig. 10 occur at a frequency of 60 Hz. This series can then be divided into five sub-series S₁, S₂' S₃, S₄, S₅. The images of the sub-series S₁ are converted in the manner as shown in Fig. 4 into the series DS₁ of difference images (system of motion vectors Q1,6, Q6,11, Q11,16, ...). The images of the sub-series S₂ are converted in the same way as is shown in Fig. 5 into the series DS₂ of difference images (system of motion vectors
Q1,6, -
Q1,6, ...). The images of the sub-series S₃ are converted in the manner as shown in Fig. 6 into the series DS₃ of difference images (system of motion vectors
Q1,6, -
Q1,6,
Q1,6,
Q6,11, ...). The images of the sub-series S₄ are converted in the manner as shown in Fig. 7 into the series DB₄ of difference images (system of motion vectors
Q1,6,-
Q1,6,
Q1,6, -
Q6,11, ...). Finally the images of sub-series S₅ are converted into a series DS₅ of difference images in the manner as shown in Fig. 7 and starting from the images in the series S₁ and S₄. All this is shown diagrammatically in Fig. 10. More particularly, each arrow starts at an image by means of which a prediction image is calculated for the image where the arrow head of the relevant arrow ends, all this while taking the correct system of motion vectors into account. By selecting only the difference images of, for example the series DS₁, DS₂, DS₃ and DS₄ of the series of difference images thus obtained and by displaying them with mutually equal intervals, an image sequence of 50 Hz is obtained. By providing a display apparatus according to Fig. 9 with a selection circuit 4 and by ordering the video images on the disc and recording them in the manner as described above with reference to Fig. 10, the discs can be used in the so-called 50 Hz countries as well as in the so-called 60 Hz countries and the display apparatus can be simply made suitable for use in these different countries.
[0028] It is to be noted that it has been assumed in the foregoing that the motions in the image are linear. Consequently it is sufficient to calculate systems of "main" motion vectors for the images in the sub-series Si. The motion vectors of the images in the other sub-series can then be obtained by taking a proportional part of these main motion vectors. However, it is alternatively possible to calculate the actual motion vectors for each image instead of taking the proportional part of the main motion vectors.
1. A method of transmitting a series of images of a full motion video scene in which
each image is converted by means of an encoding algorithm into an image data block
comprising so much digital information that each pixel of the image can be reconstructed,
characterized in that the images of the series are subjected to a hierarchic encoding
process in which the original series of images is considered as a number of interleaved
sub-series having an increasing ranking order and in which images from one or more
sub-series having a lower ranking order are considered for encoding an image of a
sub-series, and in that a packet header indicating the ranking order of the sub-series
with which the corresponding image is associated is added to each image data block.
2. An apparatus for displaying a full motion video scene whose composite images are each
converted into image data blocks in accordance with the method as claimed in Claim
1, which apparatus is adapted to receive the image data blocks, to store for each
pixel a codeword corresponding to the color of said pixel, to reproduce the stored
codewords in the sequence of desired display, to convert the reproduced codewords
into analog signals for display on a display screen, characterized in that said apparatus
is further adapted to select those image data blocks of the received image data blocks
which have predetermined packet headers, and to subject the image data blocks thus
selected to a hierarchic decoding process for generating said codewords.
3. An optically readable disc on which a series of images of a full motion video scene
is stored in the form of a series of image data blocks obtained in accordance with
the method as claimed in Claim 1.